By Sabrina Aggarwal, MPH, Senior Vice President
I went a long time without checking out ChatGPT, until quite frankly, I could not anymore. To be a well-informed healthcare public relations (PR) professional, I realized I needed to educate myself on the tool everyone was talking about. The ultimate catalyst was a headline that popped up in my Instagram feed – ‘AI Has Developed a Treatment for Cancer in 30 Days and Can Predict Survival Rate’. This headline and knowing this week is National Public Health Week – led me to experiment with this artificial-intelligence (AI) chatbot and determine once and for all – is it worth all the hype? And perhaps more importantly – can it really help treat cancer or solve some of our world’s greatest public health challenges?
For those who may not know what ChatGPT is, or who need a reminder, it’s an AI tool that can generate responses to prompts. More specifically, ChatGPT, or chat generative pretrained transformer, can instantaneously write essays, answer math questions, and even write code. Technology companies abound have jumped on the AI bandwagon by developing open-source ChatGPT applications – so if one were to search for an app – you would find many. I decided to experiment with OpenAI because it was free, easy and rose to the top of my Google search.
I started by typing in basic questions into the chat to see if really worked. These included ‘How do you make Tandoori chicken?’ and ‘Who is the President of the United States’? Within seconds, a complete response appeared on screen. I was prepared for this but stunned to see it happen in real-time. Nonetheless, I knew I could get the same quick answers by Googling, so was not terribly impressed. As this year’s National Public Health Week’s theme is ‘Centering and Celebrating Cultures in Health’, I decided to see what I could do to make a difference in public health, and ChatGPT said (abbreviated), “Volunteer, advocate, donate, educate, practice health behaviors, join a public health organization and pursue a career in public health.”
Then I took it a step further by asking a philosophical question that may not be Google-able such as ‘Does God exist?’ The response was diplomatic, but also not too impressive.
The next question I asked was ‘How can AI help global health’? This time, I sat up in my seat.
AI can play a critical role in improving global health in several ways:
Disease diagnosis: AI can be trained to identify patterns and symptoms of diseases which can be used to improve early diagnosis and treatment.
Healthcare management: AI can help manage scarce resources such as doctors, nurses, and medicines by optimizing their use and predicting demand for them. This can help to reduce waste and ensure that resources are used effectively.
Disease surveillance: AI algorithms can be used to monitor and track disease outbreaks in real-time, helping public health officials to respond more quickly and effectively.
Drug discovery: AI can be used to analyze large amounts of data related to the structure and function of proteins and molecules, which can facilitate the discovery of new drugs and treatments.
Telemedicine: AI-powered telemedicine solutions can enable remove consultations and diagnosis, particularly in underserved areas where healthcare sources may be limited. Overall, AI has the potential to revolutionize global health by providing cost-effective, efficient and scalable solutions to complex health challenges.
Wow. Not only did the chatbot respond in less than two seconds, but the response was thoughtful, robust, intelligent and efficient. It seemed to summarize a collection of data points and drill it down into an easy-to-understand way, which Google certainly cannot do. I had also just seen the headline about AI and cancer treatment, so decided to look into the study to learn more about AI’s applications in healthcare.
Researchers from the University of Toronto partnered with Insilico Medicine to use an AI-powered database called AlphaFold to create a drug that could potentially treat hepatocellular carcinoma (HCC), a type of liver cancer. Their research, published in Chemical Science, successfully identified an undiscovered pathway to treat HCC. The AI tool was able to predict the structure of all proteins in the human body, leading researchers to believe AlphaFold, amongst others, could generate novel treatments to tackle diseases such as cancer with high unmet need. In fact, another study published in JAMA Network Open found that “AI was more than 80% accurate in predicting cancer patient survivor rates.” The AI tool was able to do this by analyzing doctor’s notes on more than 50,000 patients to predict survival rates – a task that would take any human, or humans, a very long time to do. Last year, the National Cancer Institute published a piece on the potential of AI in cancer diagnoses, noting that AI can analyze thousands of MRI scans to determine the stage of a tumor without any human supervision. As ChatGPT had rightfully answered in my prompt, from the few examples I researched, AI had the potential to improve cancer screening and diagnosis, aid with healthcare management and drug discovery.
As with anything, there are drawbacks, limitations and even fears – associated with ChatGPT and other forms of AI. After I excitedly told my husband about the potential of AI and ChatGPT, he said “Haven’t you read about Superintelligence? This stuff is not what it’s hyped to be”. Superintelligence, a book written by Nick Bostrom and published in 2015, warns about the consequences of AI with questions, such as, ‘What if machine brains surpass human brains in general intelligence, then this new superintelligence could become extremely powerful – possibly beyond our control?’ I hadn’t thought about this, and suddenly had visions of large computer robots, the size of King Kong, taking over cities and eventually the world.
So, I decided to ask ChatGPT what the drawbacks of AI were to global health. Its response:
AI has enormous potential to transform global health and improve healthcare outcomes. However, there are also potential drawbacks and challenges, that need to be considered, such as:
Data bias: AI algorithms are only as good as the data they are trained on. If the data used to train the algorithm is biased or incomplete, the resulting algorithm
Lack of transparency: Some AI algorithms are complex and difficult to interpret, which can make it difficult for healthcare providers to understand how decisions are being made
Privacy concerns: AI requires access to large amounts of data, which can raise privacy concerns if that data is not handled properly
Dependence on technology: AI should not be viewed as a replacement for human healthcare providers. In some cases, reliance on technology can result in a loss of personal touch or communication, which can negatively impact patient care
Cost: Developing and implementing AI technology can be expensive, which may limit its availability and accessibility in certain regions or communities
Ethical considerations: AI raises ethnical questions around issues, such as accountability, transparency, and the potential for unintended consequences. These issues need to be carefully considered to ensure that AI is being used in a responsible and ethnical manner
Critics of ChatGPT and other forms of AI agree – that not only are there things ChatGPT cannot do – but there are also instances where ChatGPT is simply wrong. The New York Times has several podcasts on ChatGPT, including this recent one Did Artificial Intelligence Just Get Too Smart in which the hosts say the technology can and has gotten math and physics questions wrong. It’s not fool proof, and where the danger lies, is if we solely rely on AI to replace work previously done by humans. As a PR practitioner, I know for example, that ChatGPT can churn out loads of draft communications materials – and I know there has been a lot of discourse about the pros and cons of tapping into this free resource.
While I won’t be turning to ChatGPT to answer my existential questions anytime soon or replace my job responsibilities, I’m very excited to use the tool to help me become a better informed and impactful communications practitioner and augment my knowledge. I’m looking forward to asking questions, such as, ‘How do monoclonal antibodies work?’ and ‘What interleukin proteins are most beneficial to the immune system?’
I encourage all of us to experiment with ChatGPT and see what you can learn to help make our world a safer, healthier place, particularly during National Public Health Week. Maybe a computer touch is needed before one can make a human touch.